COVID-19 and olfactory dysfunction: a looming wave of dementia?

Olfactory dysfunction is a hallmark symptom of COVID-19 disease resulting from the SARS-CoV-2 virus. The cause of the sudden and usually temporary anosmia that most people suffer from COVID-19 is likely entirely peripheral-inflammation and other damage caused by the virus in the sensory epithelium inside the upper recesses of the nasal cavity can damage or prevent chemicals from properly activating the olfactory sensory neurons. However, persistent olfactory dysfunction from COVID-19, in the form of hyposmia and parosmia (decreased or altered smell) may affect as many as 15 million people worldwide. This epidemic of olfactory dysfunction is thus a continuing public health concern. Mounting evidence suggests that the SARS-CoV-2 virus itself or inflammation from the immune response in the nasal sensory epithelium may invade the olfactory bulb, likely via non-neuronal transmission. COVID-19-related long-term olfactory dysfunction and early damage to olfactory and limbic brain regions suggest a pattern of degeneration similar to that seen in early stages of Alzheimer's disease, Parkinson's disease, and Lewy body dementia. Thus, long-term olfactory dysfunction coupled with cognitive and emotional disturbance from COVID-19 may be the first signs of delayed onset dementia from neurodegeneration. Few treatments are known to be effective to prevent further degeneration, but the first line of defense against degeneration may be olfactory and environmental enrichment. There is a pressing need for more research on treatments for olfactory dysfunction and longitudinal studies including cognitive and olfactory function from patients who have recovered from even mild COVID-19.NEW & NOTEWORTHY More than 15 million people worldwide experience persistent COVID-19 olfactory dysfunction, possibly caused by olfactory bulb damage. SARS-CoV-2 can cause inflammation and viral invasion of the olfactory bulb, initiating a cascade of degeneration similar to Alzheimer's disease and Lewy body disease. People who have had even mild cases of COVID-19 show signs of degeneration in cortical areas connected with the olfactory system. These data suggest a wave of post-COVID dementia in the coming decades.

INTRODUCTION

One of the most robust early symptoms of COVID-19 is chemosensory dysfunction. Estimates from the first wave of disease in 2020 reported anosmia in as many as 85% of people affected by the SARS-CoV-2 virus (1, 2), although a more thorough meta-analysis places the number at ∼77% when objective olfactory function tests are used (3). The most common form is sudden anosmia or hyposmia, but dysgeusia (taste dysfunction) and loss of chemesthesis (cold, hot, and irritation from the trigeminal nerve in the eyes, nose, and mouth) are also common (3, 4). Early reports suggested that smell recovery was as sudden as its loss (my friend recovered her sense of smell during treatment with monoclonal antibodies) and that most people recovered their sense of smell quickly. However, newer data on COVID-19 long-haulers suggest that a large percentage of people do not fully recover their sense of smell. Conservative estimates are that ∼15 million people worldwide have persistent smell dysfunction from COVID-19, known as smell long-haulers, and that about half of these, over 7 million people, report persistent parosmia—alteration of smells, usually bad smells (5).

The SARS-CoV-2 virus uses the angiotensin-converting enzyme 2 (ACE2) receptor to gain entry into cells (6). ACE2 expression is highest in the olfactory sensory epithelium as compared with other parts of the respiratory tract (7). This indicates a high viral load in the olfactory sensory epithelium in the upper recesses of the nose, next to the cribriform plate, through which the olfactory nerve enters the olfactory bulb (OB) in the brain. This gives the virus access to the brain via a transcribriform route (8), even without infection of olfactory sensory neurons in the sensory epithelium (9). Infection in the olfactory bulb may herald further changes that affect cognitive function.

Alzheimer’s disease, Parkinson’s disease, and other Lewy body disease are among many neurodegenerative diseases in which one of the early symptoms is olfactory dysfunction (10–12). Many of these diseases that begin with olfactory dysfunction end in dementia. However, loss of smell function is not yet useful as a diagnostic tool. We do not have olfactory tests that are sensitive enough to tell the difference between the various diseases and possibly normal decreases in olfactory ability with aging (13). (It should be noted that although olfactory dysfunction is very common in these diseases, not all patients experience smell loss and not all patients who experience smell loss progress to dementia.)

Neurodegeneration patterns in Alzheimer’s disease and Lewy body disease, including Parkinson’s disease, often begin in the OB (10), and some have hypothesized that OB damage from viral invasion or an inflammatory response may be the catalyst for such degeneration in vulnerable individuals (14).

The OB is an integral part of the limbic system; the entire set of structures including the olfactory, amygdala, and hippocampal areas were for a time referred to as the rhinencephalon or smell brain (15). A schematic of olfactory, limbic, and hippocampal connections is summarized in Fig. 1. Olfactory bulbectomy causes spatial disorientation (16, 17), and Dr. Lucia Jacobs at U.C. Berkeley has argued convincingly that the OB evolved, and still functions, as part of the navigation network (18). All of this means that the OB is involved in much more than smelling. It is involved in the sense of place, memory, context, emotion, reward, and many other processes. This may explain the conservation of the OB’s anatomical connections with other systems across mammalian species (19).

Figure 1.

Schematic of the relevant olfactory and limbic system connections. Input comes to the OB directly from the olfactory nerve in the nose. Shapes with yellow shading receive direct input from the OB. Shapes with blue shading receive direct input from the PC. Not all connections are shown. Note that the amygdala consists of many smaller structures, the PC has both anterior and posterior sections with differing connections, the EC has medial and lateral portions with different connections with other sensory systems and the HPC, and the insula has subdivisions associated with several neural systems, including the gustatory system. Missing from this schematic are connections with other cortical systems, the thalamus and brainstem neuromodulatory, sensory and motor nuclei. ACC, anterior cingulate cortex; amyg, amygdala; AON, anterior olfactory nucleus; EC, entorhinal cortex; HPC, hippocampus; INS, insula; OB, olfactory bulb; OFC, orbitofrontal cortex; OSNs, olfactory sensory neurons in the nose; PC, piriform cortex; PHC, parahippocampal cortex; PRC, perirhinal cortex; TuS, tubular striatum (olfactory tubercle).

OLFACTORY BULB INFLUENCE ON THE LIMBIC SYSTEM

The OB sends monosynaptic inputs to all parts of the olfactory system (anterior olfactory nucleus, piriform cortex, and olfactory tubercle). The OB also connects monosynaptically to the amygdala, lateral and medial entorhinal cortex, taenia tecta, anterior hippocampal continuation, indusium grisium, and insula. Inputs to the OB from the brain include all of the olfactory areas, the nucleus of the lateral olfactory tract, plus lateral entorhinal cortex, ventral hippocampus, and amygdala. All neuromodulatory systems also send projections to the OB: locus coeruleus, basal forebrain, raphe nuclei, ventral tegmentum, and substantia nigra (20–23). Anatomically and physiologically, the OB is mirrored in structure by the dorsal thalamus and thalamic reticular nucleus (24).

Neuromodulation and other purveyors of cognitive context modify all types of OB activity—both amplifying and desynchronizing coordinated neural activity in the form of gamma oscillations (40–110 Hz) of the local field potential and the trade-off between odor discrimination and categorization (25–31).

Lesions of the OB cause a robust state of behavioral dysfunction that mimics aspects of depression in humans. In the late 1970s and 1980s, the olfactory bulbectomy model for depression was developed and tested. In rats, mice, and rhesus and vervet monkeys complete bilateral removal of the OBs produces depression-linked anxiety (32–36; Albert Sattin, personal communication). OB lesioned rodents have provided a robust model for testing new antidepressants. In particular, the time course of recovery from depressive behavior under antidepressant treatment provides a valid model for predicting the speed of action of antidepressants (37).

The OB influences hypothalamic activity and thus neuroendocrine function. OB mitral cells project to the suprachiasmatic nucleus and other parts of the hypothalamus. Olfactory bulbectomy alters hypothalamic pathways involved in prolactin, growth hormone, and glucocorticoid production (38, 39). Interestingly, olfactory bulbectomy in prepubescent male rats releases seasonal influences on gonadal function in a species that is normally insensitive to changes in daylength (40). Removal of the OBs causes widespread changes in immune function, which may contribute to changes in monoaminergic neurotransmission and depression (41). The OBs also influence immune function in Siberian hamsters (42).

The OB regulates limbic system neural activity. Mitral and tufted cells in the OB fire about ten times faster than other cortical neurons, 10–20 Hz on average but up to 60 Hz in a single burst within a sniff. About half of these neurons are driven by respiration (43), and this respiratory signal shows up in every cortical area that has been studied but especially in olfactory and other limbic areas (44–47). This means that the OB provides constant high frequency excitatory input to most of the limbic system with every inhalation. Removing that background excitation may be what causes the limbic system to function outside of its normal dynamic range in the olfactory bulbectomy model. Why this results in anxious and anhedonic behaviors in rats, mice, and monkeys is a puzzle.

Dr. Walter J. Freeman showed that OB input affects the dynamic range of downstream cortical areas, and this constant background supports γ oscillatory activity produced locally in the piriform cortex and entorhinal cortex (48, 49). OB γ oscillations of the local field potential, representative of coordinated neural activity produced with each sniff, have been linked recently to the mechanism for maintaining limbic tone. Artificial enhancement of OB-driven γ oscillations in piriform cortex reduces effects of depression in mice (50).

OLFACTORY BULB INSULTS MAY INITIATE CASCADES OF DAMAGE IN THE LIMBIC SYSTEM

Excitatory input to limbic areas is not the only important factor for limbic system health. Degenerative processes that begin in the OB may trigger a cascade of degeneration that results in dementia as hippocampal and other cortical areas become involved over time. Most of the evidence supporting such a mechanism at work in humans is circumstantial, but preclinical animal models offer evidence to support the idea neurodegenerative changes beginning in the OB may radiate into the limbic system (14).

Misfolded α-synuclein has been implicated in Lewy body dementia associated with Parkinson’s disease. Injection of α-synuclein preformed fibrils into the OB of mice, to seed pathological formations of endogenous α-synuclein, results in the spreading of pathological α-synuclein in the connected olfactory areas as well as other parts of the brain (51). One scenario that has been proposed is that Lewy body disease and τ protein misfolding in Alzheimer’s disease is a prion-like change that induces misfolding in wider areas over time (52). The key to this argument is that an initial insult from pathogen invasion or oxidative stress resulting in inflammation induces a change in protein folding, which triggers the chain reaction.

The idea that viral invasion of the central nervous system (CNS) can be a trigger for neurodegeneration resulting in later neurological deficit is not a new one. The 1918 pandemic produced a wave of postencephalitic Parkinson’s disease caused by delayed but fast degeneration of dopaminergic neurons in the substantia nigra. An opinion paper early in the pandemic alerted us to the possibility of a delayed parkinsonian-like wave associated with olfactory neurodegeneration from COVID-19 (53). Flu pandemics over the past 70 years have provided evidence that inflammatory effects on fetal brain development from maternal infection are correlated with schizophrenia that emerges typically during vulnerable late adolescence (54). More recently, data from a large population from the Danish National Patient Registry have shown that influenza contributes to a 70% higher risk of developing Parkinson’s disease 10 years later (55).

The OB is particularly vulnerable to infection via the cribriform plate. Some viruses can infect olfactory sensory neurons directly, including many strains of influenza, herpesvirus, poliovirus, and West Nile virus (56). The transcribriform route is one method used to bypass the blood-brain-barrier for drug delivery (8). One example is nasal zolmitriptan used to treat migraines. Nasal application diffuses the drug easily into the CNS (57). Even if pathogens do not directly infect olfactory sensory neurons, they can cross the blood-brain-barrier in the cerebrospinal fluid (CSF) within the fascicles of the olfactory and nerve in the upper recesses of the nasal cavity or via infection of the glial olfactory ensheathing cells that surround these channels (58).

The OB appears to have evolved an immune response to counteract transcribriform and sensory neuron infection (56). Innate cytokines produced within the olfactory ensheathing cells and other parts of the early olfactory pathway are upregulated in response to viral insult. OB microglia proliferate and T cells infiltrate the neural tissue. These mechanisms allow the OB to maintain vigilance against viral intruders, but if the inflammatory process does not resolve quickly, neurons can be damaged (59). Damage to OB neurons may be an initiating factor in the degeneration described at the beginning of this section.

CAUSES OF COVID-19 ANOSMIA

In light of the possibility for progressive OB damage, it is important to examine the likelihood for such damage due to infection by the SARS-CoV-2 virus. COVID-19 anosmia led many scientists to fear the worst, anosmia caused by viral invasion in the OB. The quickest most damaging way to infect the OB would be via a transneuronal route in the olfactory sensory neurons (OSNs), but fortunately this is likely not a significant factor in either anosmia or viral invasion. OSNs do not express the angiotensin converting enzyme 2 (ACE2) receptor or the transmembrane proteases, such as Transmembrane protease, serine 2 (TMPRSS2), necessary for SARS-CoV-2 to enter a cell (9, 60). Expression of ACE2 in sustentacular cells in the olfactory sensory epithelium is hundreds of times higher than in other parts of the respiratory epithelium (61). This could produce an extremely high viral load in the upper recesses of the nasal cavity, adjacent to the cribriform plate (7, 62). Other non-neuronal cells in the olfactory sensory epithelium that are heavily impacted by SARS-CoV-2 infection are the Bowman’s glands, basal cells (the stem cells that give rise to new OSNs) and the olfactory ensheathing cells. Infection in all of these can produce sudden anosmia, the cause of which is likely peripheral. The fast-onset anosmia in COVID-19 caused by inflammation thus makes sense considering its short latency and, in most cases, its rapid remission.

Long-term hyposmia, anosmia, or parosmia may involve peripheral or central mechanisms. Peripheral causes may include cell death or damage to the non-neuronal cells in the sensory epithelium. Although these cells regenerate throughout life, extensive damage may impair the sensory epithelium’s capacity to regenerate, especially if basal cells are impacted (63, 64). A recent report suggests that one cause of persistent anosmia from COVID-19 may be downregulation of olfactory receptor genes (65).

Even if the sensory epithelium is not permanently damaged, if OSNs are damaged in large numbers, the sensory neurons may need weeks or months to regenerate and sort out their connections to the OB glomeruli, resulting in long-lasting parosmia, or they may just wire differently, requiring new experience for the brain to learn the patterns for “new” odor inputs (66). All of these possibilities may be relatively benign for future brain health, even though they can be devastating for quality of life.

EVIDENCE OF DAMAGE TO THE OLFACTORY BULB

Although the SARS-CoV-2 virus does not enter the OSNs, the virus can enter the OBs and the rest of the brain by means other than the transneuronal route. The ensheathing cells that cover the sensory nerve bundles that travel from the nose to the OB have the appropriate receptors that allow invasion by the SARS-CoV-2 virus. In addition, weakened capillary walls, caused by the viral infection, would allow viral entry into the extracellular space. Once the virus invades the OB by either the transcribriform or vascular route, it can infect the many types of glial cells expressing ACE2 receptors (67). A cascade of inflammatory responses among microglial and astrocytic activations can cause neurodegeneration and synapse loss, the neurological consequences of which would be expected to vary, depending on the brain region. There may be many more types of inflammatory effects, but cataloguing and considering all the downstream mediators and cell types is beyond scope of this review.

Recent studies report changes in the OB due to inflammation from COVID-19. Changes in gene expression are reported in the OB and amygdala of patients who died from COVID-19 as compared with deaths from other causes (68). In the COVID-19 cases, expression of OB genes associated with microvascular damage and inflammation were many times higher than those in the amygdala. Neuroimaging shows evidence of postviral inflammation in the OB and olfactory tract, and autopsy studies point at an activated immune response in the OB, including sterile inflammation of the OB due to the nearby inflammation of the olfactory epithelium (69).

There is evidence that the SARS-CoV-2 virus does enter the OB selectively relative to nearby parts of the olfactory system. In a sample of 16 postmortem brains from COVID-related deaths, viral DNA and positive RT-PCR for the virus were found in the OBs from eight of the brains. The authors concluded that the invasion was likely via the blood supply that the OB shares with the olfactory epithelium and not systemic blood supply, because there was no evidence of infection in other closely connected and nearby olfactory areas (70).

Other neural routes in the vicinity of the OB are the terminal nerve (nervus terminalis or Cranial Nerve 0) and the trigeminal nerve. Both have ACE2 receptors and dense ramification in the nasal cavity, in the olfactory sensory epithelium and around the cribriform plate. The terminal nerve may bypass the OB and project to the hypothalamus and other parts of the brain posterior caudal to the OB (71, 72). There is some evidence that GnRH neurons are compromised by SARS-CoV-2, suggesting involvement of the terminal nerve (73). Although the terminal nerve does not appear to synapse within the OB, some fibers do make contact with the CSF in the OB (74). So, the terminal nerve could be infected along with infection of the OB via the vascular and transcribriform route.

The trigeminal nerve is another route by which the virus could enter the brain. Nerve endings in the eyes, nose and mouth produce sensations of hot, cold and irritation from chemicals, like the cold feeling in the nose with mint toothpaste or burning in the eyes and nose while cutting onions. The cell bodies for this nerve are in the brainstem and most fibers are in the periphery in the face. In rats, the sensory and neuromodulatory free nerve ending collaterals travel with the olfactory nerve bundle into the OB, where they can release Substance P and Calcitonin gene-related peptide (CGRP) to modulate activity of mitral and tufted cells (75). Several pieces of evidence support SARS-CoV-2 invasion of the trigeminal pathway. The trigeminal nerve neurons express ACE2 and TMPRSS2 (76). Virus has been detected in all branches of the trigeminal nerve (77). Degeneration of the trigeminal nerve has been observed in autopsies from COVID-19 deaths (78). Finally, chemesthetic dysfunction is common in COVID-19 (2, 79).

Changes to the OB could also be the result of long-term loss of sensory input in COVID-19 anosmia. There are several reports of decreased OB volume in patients with long COVID associated with olfactory dysfunction (80). This finding mirrors the decreased OB volume seen with posttrauma and postinfection smell dysfunction, particularly parosmia (81). Decreases in OB volume could be followed by changes in downstream areas that receive OB input.

Damage to the OB and progressively to the broader limbic system would predict memory and emotional deficits and, importantly, degeneration of areas connected to the OB and piriform cortex. Neuropsychological assessments 6–9 mo after severe, moderate, or mild COVID-19 disease showed multiple cognitive and emotional deficits in all groups (82). For those with more severe infection, there was a correlation between continued olfactory dysfunction and severity of cognitive and emotional dysfunction.

A recent study using data from the UK Biobank examined brain scans and cognitive measures from subjects who were tested before the pandemic and more recently (83). Subjects who had had mild COVID-19 (not requiring hospitalization) in the interim, compared with age-matched controls, showed cortical thinning and/or degenerative changes in areas connected to the OB, olfactory tubercle [tubular striatum (84)], anterior olfactory nucleus, and piriform cortex—the parahippocampal/perirhinal cortex, entorhinal cortex, hippocampus, insula, orbitofrontal cortex, anterior cingulate cortex, and amygdala. Overall brain volume was decreased and CSF volume increased in the COVID-19 group relative to controls, indicating diffuse loss of gray matter from COVID-19. Cognitive tests showed that COVID-19 associated decline in executive function, correlated with changes in crus II of the cerebellum, known to be involved in olfactory processing (85). The average time since the subjects’ COVID-19 diagnosis was ∼5 mo.

THE POSSIBLE ROLE OF NEUROPLASTICITY IN RECOVERY

Enrichment studies show that environmental and behavioral factors can have a powerful influence on neural architecture and cognitive function. In the 1960s, Marian Diamond, David Krech, and Mark Rosenzweig first described effects of environmental enrichment on brain health and cognition. Rats housed in complex environments with changing stimuli, toys, space, apparatus to climb and exercise on, and abundant social interactions with cage mates have thicker cortex and better cognitive ability than rats living impoverished solitary lives in standard, nonenriched housing (86). Cortical thickening is driven by proliferation of glial cells (87). More recent work has identified astrocytes as important functional units in synapse formation and maintenance (88). Although there is little evidence for adult neurogenesis in neocortex, there is ample evidence for increase of adult born glial cells of all types. Proliferation of synapses means proliferation of astrocytes and other glial cells.

Adult hippocampal neurogenesis is elevated in rats that exercise and explore space frequently as compared with those living solitary and uneventful lives in standard laboratory cages (89). There is reason to be skeptical that humans enjoy the same level of adult hippocampal neurogenesis as rats and mice, but recent research and discussions once again argue that humans may also grow new hippocampal neurons after the early developmental stage (90). There is growing evidence that environment and daily habits, such as learning or exercise may contribute to the health and possibly growth of the hippocampus in adulthood (91, 92).

Smell training may help people recover their sense of smell but only if there is still an intact and functional sensory epithelium. We know from studies in rats and mice that enrichment with odors promotes growth of new GABAergic granule cells in the OB and increases olfactory ability (93–96). There is some evidence that humans may also grow new OB neurons throughout life (97, 98). However, there is also some doubt that OB neurogenesis is functional in humans (99). Even without neurogenesis, with a functional sensory epithelium, odor enrichment may give new sensory neurons more chance to grow and make connections with the OB. New inputs to existing neurons will help them survive and promote proliferation of synapses and glial cells. Smell training for six or twelve weeks or more for patients with long term hyposmia helps to increase olfactory ability and cortical thickness in areas that were thinner before training in these patients relative to controls (100, 101). Smell training also increases olfactory ability and cortical thickness even in people with normal olfactory ability (102) (visual training did not increase cortical thickness in this study, which is an interesting control group). Smell training may work for many to treat postviral anosmia or hyposmia, but it can take months and the success rate ranges from 25% to 50%, and the method has only been tested in small cohorts (103).

SUMMARY AND LOOKING FORWARD

Olfactory bulbs are particularly vulnerable to insult from infection in the nasal cavity. They can be infected by a transneuronal (via the olfactory or trigeminal nerve) or via the transcribriform route, through the olfactory nerve fascicles, CSF, or terminal nerve. Another possible route is the vasculature in the sensory epithelium in the nose and the olfactory bulb. Presently, there is evidence for all of these routes of infection, except for direct transneuronal infection of the olfactory sensory neurons. Damage to the olfactory bulb and other early olfactory areas is evidenced by presence of virus in OBs at autopsy, OB inflammation (including sterile inflammation), microvascular damage in the OB, reduced OB volume, and reduced volume of cortical areas connected to the olfactory system. These changes could explain the persistent hyposmia and parosmia experience by many millions of people worldwide. Early evidence suggests cognitive symptoms due to degenerative processes triggered by COVID-19. All of these factors support the possibility of long-term cognitive damage leading to dementia in many of these cases.

Our best current tool to fight degeneration of olfactory and other parts of the cerebral cortex may be olfactory and environmental enrichment—smelling multiple familiar nontoxic odorants daily in a long therapeutic course (smell training) for olfactory function, plus physical exercise and cognitive training for global brain health. Smell training shows some promise (103, 104), but it still awaits large, controlled studies that prove efficacy. There are no other treatments available. These are all relatively simplistic solutions, but we have very few other therapies that have been shown to produce positive changes in brains and cognitive ability. Fortunately, these activities are cheap and easy compared to late-stage dementia care.

Much more research is needed to develop treatments for chemosensory dysfunction. We need good public health support, more money to support research on recovery from smell disorders, and public education on all of these factors. Given the immunological effects evident in the olfactory bulbs, research into therapeutic approaches that target immunological players in the CNS could be helpful. Large longitudinal cognitive and neurophysiological studies including people of all ages who have had even mild cases of COVID-19 are essential and may also shed light on the role of olfactory neuroinflammation and degeneration associated with other diseases. Clinicians should continue to monitor patients who have had COVID-19 for continued or progressive smell dysfunction and related cognitive and emotional changes. The Alzheimer’s Association has teamed up with an international consortium to monitor signs of dementia over the long term as a result of COVID-19 disease (105), but larger efforts that include emphasis on olfactory dysfunction are necessary.

CODA

This entire story is based on a hunch that comes from my deep knowledge of olfaction and its role in limbic system health. The hunch is supported by confirmatory evidence. Maybe I am wrong. I hope I am wrong. There is not yet proof that infection in the OB will lead to dementia later on. However, there is enough evidence from the current pandemic and the place of the olfactory system in many diseases that result in dementia that further research is warranted. In any case, we will answer important and urgent questions by paying attention to them now. In 1920, in the wake of the Spanish Flu pandemic, we did not have the research infrastructure or technologies that we have now. Although a catastrophe on many levels, the COVID-19 pandemic presents an opportunity to improve human health. We should take advantage of this opportunity.

GRANTS

This work was funded by a University of Chicago Big Ideas Grant.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

Leslie Kay is an editor of Journal of Neurophysiology and was not involved and did not have access to information regarding the peer-review process or final disposition of this article. An alternate editor oversaw the peer-review and decision-making process for this article.

AUTHOR CONTRIBUTIONS

L.M.K. conceived and designed research; prepared figures; drafted manuscript; edited and revised manuscript; approved final version of manuscript.